AWS Organizations now automatically emits CloudTrail events to your management account whenever accounts join or leave your organization. These new events—AccountJoinedOrganization and AccountDepartedOrganization—provide security teams and cloud administrators with enhanced visibility into organizational membership changes, helping detect unauthorized activities and potential security incidents that previously could go unnoticed.
The AccountJoinedOrganization event captures how an account joined an organization (Created or Invited) and the join timestamp, while the AccountDepartedOrganization event records how an account departed —Left for accounts that departed voluntarily, Removed for accounts removed by the management account, or Cleaned for accounts that were permanently closed along with the departure timestamp.
You can leverage these events to create CloudWatch alarms or Amazon EventBridge rules for real-time notifications, enabling rapid response to suspicious organizational changes. This capability supports critical use cases including fraud detection, compliance auditing, security monitoring, and incident investigation across your AWS environment.
AWS Organizations now automatically emits CloudTrail events to your management account whenever accounts join or leave your organization. These new events—AccountJoinedOrganization and AccountDepartedOrganization—provide security teams and cloud administrators with enhanced visibility into organizational membership changes, helping detect unauthorized activities and potential security incidents that previously could go unnoticed.
The AccountJoinedOrganization event captures how an account joined an organization (Created or Invited) and the join timestamp, while the AccountDepartedOrganization event records how an account departed —Left for accounts that departed voluntarily, Removed for accounts removed by the management account, or Cleaned for accounts that were permanently closed along with the departure timestamp.
You can leverage these events to create CloudWatch alarms or Amazon EventBridge rules for real-time notifications, enabling rapid response to suspicious organizational changes. This capability supports critical use cases including fraud detection, compliance auditing, security monitoring, and incident investigation across your AWS environment.
Amazon Connect Customer now supports generative AI-powered post-contact summaries in eight additional language families: Portuguese, French, Italian, German, Spanish, Chinese, Japanese, and Korean. Post-contact summaries also now support non-US variations of English, including British English, Australian English, and other regional locales, ensuring summaries reflect locally appropriate spelling and terminology.
Generative AI-powered post-contact summaries provide agents and managers with concise, structured overviews of customer conversations across voice, chat, and email channels, eliminating the need to read full transcripts. With this expansion, organizations can automatically generate summaries in the language of the conversation, helping agents complete after-contact work faster and enabling managers to review contacts across languages. For example, a global support organization can now generate post-contact summaries for calls handled in French, German, or Japanese, giving supervisors visibility into service quality across all regions.
The newly supported languages are available in all AWS Regions where Amazon Connect Customer post-contact summaries are available. To learn more, refer to View generative AI-powered post-contact summaries in the Amazon Connect Customer Administrator Guide. To learn more about Amazon Connect Customer, visit the Amazon Connect Customer website.
Amazon Connect Customer now supports generative AI-powered post-contact summaries in eight additional language families: Portuguese, French, Italian, German, Spanish, Chinese, Japanese, and Korean. Post-contact summaries also now support non-US variations of English, including British English, Australian English, and other regional locales, ensuring summaries reflect locally appropriate spelling and terminology.
Generative AI-powered post-contact summaries provide agents and managers with concise, structured overviews of customer conversations across voice, chat, and email channels, eliminating the need to read full transcripts. With this expansion, organizations can automatically generate summaries in the language of the conversation, helping agents complete after-contact work faster and enabling managers to review contacts across languages. For example, a global support organization can now generate post-contact summaries for calls handled in French, German, or Japanese, giving supervisors visibility into service quality across all regions.
The newly supported languages are available in all AWS Regions where Amazon Connect Customer post-contact summaries are available. To learn more, refer to View generative AI-powered post-contact summaries in the Amazon Connect Customer Administrator Guide. To learn more about Amazon Connect Customer, visit the Amazon Connect Customer website.
11 startups a tener en cuenta Microsoft Build 2026
Por: Microsoft para Startups.
Desde el prototipo hasta la producción: el nuevo estándar para startups que priorizan la IA
La conversación ha cambiado. Hace un año, los fundadores acudieron a Microsoft para preguntar si debían construir con IA. Hoy en Microsoft for Startups, la pregunta que más escuchamos es cómo hacer que la IA funcione en producción, a gran escala, en sistemas empresariales reales, y ese cambio moldea todo sobre la cohorte de Build de este año.
Por primera vez en casi una década, Microsoft Build deja Seattle. El traslado a San Francisco es deliberado: aquí es donde el ecosistema de infraestructura de IA es más denso, donde las startups que redefinen las herramientas, el cálculo, la observabilidad y los datos para desarrolladores hacen su trabajo más agudo, y donde ese trabajo está más cerca de los compradores empresariales a los que necesita llegar. El evento de este año refleja un enfoque técnico más agudo que nunca, organizado en torno a las decisiones de ingeniería que en verdad determinan si la IA se lanza, escala y resiste en condiciones reales.
Las startups aquí presentadas reflejan esa agenda. No construyen pruebas de concepto. Trabajan en los problemas difíciles y poco glamurosos entre una buena idea y un sistema que en verdad funciona: cómo autenticar agentes sin exponer credenciales, cómo entender bien el código heredado para modernizarlo de manera segura, cómo saber si las herramientas de IA que adoptó el equipo de ingeniería en verdad los hacen más rápidos. Estas son las preguntas que importan en 2026.
La adopción de la IA ha seguido un arco previsible: los primeros experimentos y pilotos dieron paso a una realidad más dura, donde los modelos alucinan, los agentes se desvían, los datos son sensibles y los costes se acumulan más rápido de lo que nadie presupuestaba. Las empresas aprendieron con rapidez que enviar IA a producción requiere una clase de herramientas diferente a la de enviar a un entorno de demostración, y esa brecha ha impulsado un aumento de la inversión en la capa de infraestructura.
Las necesidades son específicas y atraviesan toda la pila. En el ámbito de los datos, eso significa bases de datos diseñadas en específico para la recuperación multimodal a gran escala y plataformas de datos sintéticos que ofrecen a los desarrolladores datos de calidad de producción sin exponer nada sensible. En el ámbito de la infraestructura, significa marcos de trabajo de cómputo que escalan cargas de trabajo de IA distribuidas sin un equipo dedicado, y plataformas de observabilidad que aportan la misma visibilidad al comportamiento de los modelos de lenguaje que los equipos de ingeniería siempre han tenido en el rendimiento de las aplicaciones.
Microsoft Build 2026 está organizada justo en torno a este terreno, para cubrir arquitecturas de generación aumentada por recuperación agéntica (RAG, por sus siglas en inglés), el despliegue de modelos optimizados para costes, Foundry IQ para un contexto listo para agentes y rutas de salida al mercado a través de Microsoft Marketplace, con todo el arco desde la tracción inicial hasta la escala empresarial en vista.
11 startups que lideran el camino en Microsoft Build 2026
Las startups que se unen a nosotros en Microsoft Build 2026 este año abarcan herramientas para desarrolladores, infraestructura de IA e IA física, y lo que comparten es una inclinación hacia resolver problemas empresariales reales en lugar de otros que suenan interesantes. Muchos son miembros del Programa Pegasus de Microsoft for Startups o respaldados por M12, y todos están disponibles directo en Microsoft Marketplace, lo que facilita que los clientes comerciales adquieran, desplieguen y escalen sus soluciones dentro del ecosistema Azure. Aquí tienen a quién deben tener en el radar.
Los equipos de ingeniería dedican demasiado tiempo a responder a incidentes en lugar de construir. El Hawkeye de NeuBird cambia esa ecuación: un ingeniero de fiabilidad de sitio agente (SER, por sus siglas en inglés) que interpreta la telemetría de toda su pila de observabilidad, diagnostica problemas y genera la resolución en minutos en lugar de horas. En 2025, los clientes la utilizaron para resolver 230.000 alertas y recuperar 12.000 horas de ingeniería.1 Socio de éxito de Microsoft ISV, miembro del Programa Pegasus y empresa respaldada por M12, NeuBird fue el primer SRE impulsado por IA generativa en llegar al Marketplace.
La premisa detrás de Replit es sencilla: cualquier persona en una organización, no solo los ingenieros, debería poder convertir una idea en software funcional. Con más de 500.000 usuarios empresariales en la plataforma, esa apuesta ha comenzado a dar sus frutos.1 La colaboración de Replit con Microsoft integra su plataforma de desarrollo agéntico con Azure Container Apps, Azure Virtual Machines y Neon Serverless Postgres, y la pone a disposición de manera directa a través de Marketplace. El resultado es la creación de aplicaciones de nivel empresarial sin el tradicional cuello de botella de la ingeniería.
Ray es el marco de trabajo de cómputo distribuido de código abierto que impulsa cargas de trabajo de IA en Uber, Spotify, Canva y decenas de otros, con más de 27 millones de descargas mensuales.1 Anyscale, la empresa detrás de Ray, se asoció con Microsoft para co-desarrollar un servicio Azure gestionado por completo y propietario, construido sobre él: un entorno de alto rendimiento para entrenamiento, inferencia y procesamiento de datos que se ejecuta de forma nativa dentro de tu entorno Azure Kubernetes Service (AKS), con facturación unificada y un rendimiento hasta 10 veces superior en comparación con Ray autogestionado.1
La mayoría de las herramientas de programación por IA ayudan a los ingenieros a escribir nuevo código. Moderne resuelve un problema diferente: los millones de líneas de código existente que aún necesitan ser entendidas, mantenidas y modernizadas. Su plataforma automatiza la refactorización a gran escala en miles de repositorios de manera simultánea, construida sobre OpenRewrite, el marco de trabajo de código abierto que el CEO de Moderne desarrolló en un inicio en Netflix y ahora está integrado en GitHub Copilot. Squarespace, Allstate y cinco de los principales bancos de Norteamérica lo utilizan para eliminar la deuda técnica a una escala que de otro modo requeriría decenas de miles de horas de ingeniería manual.
Antes de poder modernizar código heredado, primero hay que entenderlo. La plataforma Code-to-Spec de CoreStory utiliza IA agente para analizar millones de líneas de código existente y generar documentación viva que captura en automático reglas de negocio, relaciones con el sistema e intención del desarrollador. Lo que antes requería 18 meses de revisión manual ahora lleva días.1 En investigaciones publicadas de manera conjunta con Microsoft, el uso de especificaciones estructuradas de CoreStory dentro de los agentes de ingeniería de software de IA mejoró la precisión en un 51%,1 una compresión significativa de los plazos de modernización para equipos empresariales. Si se unen a nosotros en San Francisco, pasen por la experiencia Marketplace en el Microsoft Showcase en el Festival Pavilion para verla en acción.
Las herramientas de codificación por IA están por todas partes. Saber si mueven la aguja es un problema más difícil, y es el que Faros AI está diseñada para resolver. Su plataforma de inteligencia de ingeniería agrega datos de más de 100 herramientas, entre ellas GitHub Copilot, para brindar a los líderes de ingeniería una fuente única de verdad sobre productividad, entrega y ROI de IA. Nombrado Socio del Año Microsoft para Startups 2025 entre más de 4.600 nominaciones,1 Faros está disponible en el Marketplace con compras elegibles para beneficios de Azure, por lo que los clientes pueden aplicar el 100% de la compra a su compromiso de gasto en la nube. El equipo de Faros también estará presente en la experiencia Marketplace en el Microsoft Showcase y el Pabellón del Festival.
La mayoría de los agentes de IA fallan en producción no porque los modelos no sean capaces, sino porque no pueden actuar de manera segura sobre los sistemas empresariales que necesitan tocar. Arcade.dev es el protocolo de ejecución del protocolo de contexto del modelo (MCP, por sus siglas en inglés) que resuelve esto: proporciona la infraestructura de autorización, fiabilidad y gobernanza que permite a los agentes realizar acciones reales en Microsoft 365, Github, Teams, Salesforce, Jira y cientos de otras herramientas empresariales, sin exponer las credenciales al modelo. Es la capa de control entre agentes capaces y los sistemas a los que necesitan llegar.
La brecha entre un prototipo funcional y un robot listo para producción es todavía un desafío definitorio, ya que las herramientas, marcos y sistemas necesarios para hacer que los robots sean inteligentes nunca fueron diseñados para funcionar juntos. General Robotics construye una red de inteligencia unificada para IA física: una plataforma nativa en la nube para componer habilidades de IA para percepción, planificación y acción que puede simularse y desplegarse más rápido en cualquier factor de forma robótico. Su plataforma GRID hace que la inteligencia robótica sea más accesible, lo que brinda acceso API-first para los desarrolladores y una experiencia centrada en el agente para los operadores robots.
Las aplicaciones de IA multimodal necesitan una base de datos diseñada para el trabajo, y la mayoría no lo estaban. LanceDB es un refugio lacustre de código abierto, nativo de IA, diseñado para búsqueda vectorial a mil millones de texto, imágenes, vídeo y audio, con una arquitectura de separación cómputo-almacenamiento que se ejecuta de forma nativa en Azure Storage. Se integra con LangChain, LlamaIndex y DuckDB, y empresas como Midjourney, Runway y Character.ai lo ejecutan en producción.1 LanceDB recaudó una Serie A de 30 millones de dólares en 2025 para acelerar su plataforma empresarial.
Lanzar un modelo de IA es una cosa. Saber cómo se comporta en verdad en producción es otra cosa. Arize AI es la plataforma de observabilidad y evaluación creada para ese segundo problema, que ofrece a los equipos de ingeniería la misma visibilidad sobre el comportamiento de los modelos de lenguaje que siempre han tenido en el rendimiento de las aplicaciones. Con más de dos millones de descargas mensuales de su biblioteca de código abierto Arize Phoenix1 y profundas integraciones con Microsoft Foundry y Azure AI Studio, Arize es la capa de monitorización sobre la que las empresas construyen. M12, el fondo de riesgo de Microsoft, participó en la Serie C de la compañía, valorada en 70 millones de dólares.
El problema de los datos en la IA empresarial es específico: los datos más útiles para el entrenamiento y las pruebas son también los datos más restringidos. Tonic AI resuelve esa tensión con datos sintéticos de alta fidelidad que reflejan la calidad de producción sin tocar nada sensible, instalándose directo en el inquilino Azure del cliente con soporte para Azure OpenAI, Microsoft Fabric y Azure SQL. Miembro del Programa Pegasus con compras elegibles para beneficios en Azure a través de Marketplace, Tonic ya ayuda a equipos de Comcast, eBay y UnitedHealthcare a avanzar más rápido sin comprometer el cumplimiento normativo.
La capa de infraestructura de IA se construye ahora mismo, y las startups que aparecen aquí realizan algunos de los trabajos más importantes en ella. Ya sea que nos acompañen en San Francisco o que vean el evento en línea, Microsoft Build 2026 es donde ocurren esas conversaciones.
¿Quieren conocer a las startups detrás de la historia? Únanse a nosotros en Dev Your Own Way, organizado por Microsoft for Startups el 2 de junio de 2026 en San Francisco, California. Conectarán con fundadores, asistentes a Microsoft Build, expertos técnicos y líderes de todo el ecosistema de startups y desarrolladores.
AWS now offers Claude Opus 4.8 — Anthropic’s most capable generally available model to date — delivering meaningful advances across agentic coding, professional knowledge work, and long-running autonomous tasks for developers and enterprises building production AI applications.
Claude Opus 4.8 can perform longer autonomous runs, deeper reasoning, and consistency to be trusted with production work. For coding, the Opus 4.8 reads codebases like an engineer, plans before it edits, and holds context across long sessions in real repositories. For agentic tasks, it is better at finding paths around obstacles instead of stalling, recovering from its own errors, and knowing when to ask for help versus when to keep going. For knowledge work, it better synthesizes across long documents and complex sources, self-checks its output, and delivers structured deliverables that hold up to review.
Customers have two ways to access Claude Opus 4.8: Amazon Bedrock and Claude Platform on AWS.
Amazon Bedrock keeps your data within AWS infrastructure and provides access to Claude Opus 4.8 through a unified service with AWS-managed features like Guardrails, Knowledge Bases, and regional data residency. To learn more, see Amazon Bedrock documentation and regional availability..
Claude Platform on AWS gives you direct access to Anthropic’s native platform experience and capabilities via the AWS Console. Build, test, and deploy with the same APIs, features, and console experience you’d get working with Anthropic directly, unified with AWS billing and authentication. To get started, see the Claude Platform on AWS documentation
AWS now offers Claude Opus 4.8 — Anthropic’s most capable generally available model to date — delivering meaningful advances across agentic coding, professional knowledge work, and long-running autonomous tasks for developers and enterprises building production AI applications.
Claude Opus 4.8 can perform longer autonomous runs, deeper reasoning, and consistency to be trusted with production work. For coding, the Opus 4.8 reads codebases like an engineer, plans before it edits, and holds context across long sessions in real repositories. For agentic tasks, it is better at finding paths around obstacles instead of stalling, recovering from its own errors, and knowing when to ask for help versus when to keep going. For knowledge work, it better synthesizes across long documents and complex sources, self-checks its output, and delivers structured deliverables that hold up to review.
Customers have two ways to access Claude Opus 4.8: Amazon Bedrock and Claude Platform on AWS.
Amazon Bedrock keeps your data within AWS infrastructure and provides access to Claude Opus 4.8 through a unified service with AWS-managed features like Guardrails, Knowledge Bases, and regional data residency. To learn more, see Amazon Bedrock documentation and regional availability..
Claude Platform on AWS gives you direct access to Anthropic’s native platform experience and capabilities via the AWS Console. Build, test, and deploy with the same APIs, features, and console experience you’d get working with Anthropic directly, unified with AWS billing and authentication. To get started, see the Claude Platform on AWS documentation
Amazon DynamoDB Streams now supports AWS PrivateLink for FIPS (Federal Information Processing Standard) endpoints in AWS GovCloud (US) Regions. DynamoDB Streams captures time-ordered sequences of item-level modifications in DynamoDB tables, enabling real-time data processing and event-driven architectures. This enhancement allows government agencies and organizations with federal compliance requirements to establish private connectivity between their VPCs and DynamoDB Streams FIPS endpoints without exposing traffic to the public internet.
This capability helps customers meet strict federal compliance and regulatory requirements while simplifying their network architecture. By keeping all traffic within the AWS network infrastructure, organizations can securely process real-time data streams, implement compliant change data capture (CDC) solutions, and build event-driven architectures that adhere to federal security standards. Government agencies operating in GovCloud regions can now leverage DynamoDB Streams for secure data streaming applications while maintaining the enhanced security and privacy that AWS PrivateLink provides.
AWS PrivateLink support for DynamoDB Streams FIPS endpoints is available in AWS GovCloud (US-East) and AWS GovCloud (US-West) Regions, as well as US East (N. Virginia), US East (Ohio), US West (N. California), US West (Oregon), Canada (Central), and Canada West (Calgary).
Amazon DynamoDB Streams now supports AWS PrivateLink for FIPS (Federal Information Processing Standard) endpoints in AWS GovCloud (US) Regions. DynamoDB Streams captures time-ordered sequences of item-level modifications in DynamoDB tables, enabling real-time data processing and event-driven architectures. This enhancement allows government agencies and organizations with federal compliance requirements to establish private connectivity between their VPCs and DynamoDB Streams FIPS endpoints without exposing traffic to the public internet.
This capability helps customers meet strict federal compliance and regulatory requirements while simplifying their network architecture. By keeping all traffic within the AWS network infrastructure, organizations can securely process real-time data streams, implement compliant change data capture (CDC) solutions, and build event-driven architectures that adhere to federal security standards. Government agencies operating in GovCloud regions can now leverage DynamoDB Streams for secure data streaming applications while maintaining the enhanced security and privacy that AWS PrivateLink provides.
AWS PrivateLink support for DynamoDB Streams FIPS endpoints is available in AWS GovCloud (US-East) and AWS GovCloud (US-West) Regions, as well as US East (N. Virginia), US East (Ohio), US West (N. California), US West (Oregon), Canada (Central), and Canada West (Calgary).
To learn more, visit the Amazon DynamoDB Streams PrivateLink documentation and the AWS PrivateLink page.
We are pleased to announce general availability of Amazon EC2 P4de instances in Asia Pacific (Tokyo) on SageMaker notebook instances.
Amazon EC2 P4de instances are powered by 8 NVIDIA A100 GPUs with 80GB high-performance HBM2e GPU memory, 2X higher than the GPUs in our current P4d instances. The new P4de instances provide a total of 640GB of GPU memory, which provide up to 60% better ML training performance along with 20% lower cost to train when compared to P4d instances. The improved performance will allow customers to reduce model training times and accelerate time to market. Increased GPU memory on P4de will also benefit workloads that need to train on large datasets of high-resolution data.
We are pleased to announce general availability of Amazon EC2 P4de instances in Asia Pacific (Tokyo) on SageMaker notebook instances.
Amazon EC2 P4de instances are powered by 8 NVIDIA A100 GPUs with 80GB high-performance HBM2e GPU memory, 2X higher than the GPUs in our current P4d instances. The new P4de instances provide a total of 640GB of GPU memory, which provide up to 60% better ML training performance along with 20% lower cost to train when compared to P4d instances. The improved performance will allow customers to reduce model training times and accelerate time to market. Increased GPU memory on P4de will also benefit workloads that need to train on large datasets of high-resolution data.
Visit developer guides for instructions on setting up and using JupyterLab and CodeEditor applications on SageMaker Studio and SageMaker notebook instances.
We are pleased to announce general availability of Amazon EC2 P5.48xl instances in Asia Pacific (Tokyo) on SageMaker notebook instances.
Amazon EC2 P5.48xl instances are powered by NVIDIA H100 Tensor Core GPUs and deliver high performance in Amazon EC2 for deep learning (DL) and high performance computing (HPC) applications. They help you accelerate your time to solution by up to 4x compared to previous-generation GPU-based EC2 instances, and reduce cost to train ML models by up to 40%. Customers can use P5 instances for training and deploying complex large language models (LLMs) and diffusion models powering generative AI applications. These applications include question answering, code generation, video and image generation, and speech recognition.
We are pleased to announce general availability of Amazon EC2 P5.48xl instances in Asia Pacific (Tokyo) on SageMaker notebook instances.
Amazon EC2 P5.48xl instances are powered by NVIDIA H100 Tensor Core GPUs and deliver high performance in Amazon EC2 for deep learning (DL) and high performance computing (HPC) applications. They help you accelerate your time to solution by up to 4x compared to previous-generation GPU-based EC2 instances, and reduce cost to train ML models by up to 40%. Customers can use P5 instances for training and deploying complex large language models (LLMs) and diffusion models powering generative AI applications. These applications include question answering, code generation, video and image generation, and speech recognition.
Visit developer guides for instructions on setting up and using JupyterLab and CodeEditor applications on SageMaker Studio and SageMaker notebook instances.
Amazon Bedrock is a fully managed service that provides secure, enterprise-grade access to high-performing foundation models from leading AI companies, enabling you to build and scale generative AI applications. Amazon Bedrock customers can now view inference quotas for the bedrock-mantle endpoint through AWS Service Quotas. This gives customers a familiar, consistent way to track limits for this endpoint, the same way they already do for the bedrock-runtime endpoint and other AWS services, and gives them clear visibility into the limits that apply to their workloads.
The bedrock-mantle endpoint supports the OpenAI Responses API, OpenAI Chat Completions API, and the Anthropic Messages API, letting customers run existing OpenAI or Anthropic based applications on Amazon Bedrock with minimal code changes. AWS Service Quotas now exposes per-model input-tokens-per-minute and output-tokens-per-minute quotas for supported models on the endpoint.
With this launch, customers gain visibility into how much limits they have on the bedrock-mantle endpoint and can proactively plan for production scale. To get started, open the AWS Service Quotas console, choose Amazon Bedrock, and search for «Bedrock Mantle» to view your current quotas. To request an increase to any of these quotas, follow the standard Amazon Bedrock limit increase process. Service Quotas support for the bedrock-mantle endpoint is available in all AWS Regions where the endpoint is offered: US East (N. Virginia, Ohio), US West (Oregon), Asia Pacific (Mumbai, Tokyo, Sydney, Jakarta), Europe (Frankfurt, Ireland, London, Milan, Stockholm), and South America (São Paulo). To learn more, see Quotas for Amazon Bedrock.
Amazon Bedrock is a fully managed service that provides secure, enterprise-grade access to high-performing foundation models from leading AI companies, enabling you to build and scale generative AI applications. Amazon Bedrock customers can now view inference quotas for the bedrock-mantle endpoint through AWS Service Quotas. This gives customers a familiar, consistent way to track limits for this endpoint, the same way they already do for the bedrock-runtime endpoint and other AWS services, and gives them clear visibility into the limits that apply to their workloads. The bedrock-mantle endpoint supports the OpenAI Responses API, OpenAI Chat Completions API, and the Anthropic Messages API, letting customers run existing OpenAI or Anthropic based applications on Amazon Bedrock with minimal code changes. AWS Service Quotas now exposes per-model input-tokens-per-minute and output-tokens-per-minute quotas for supported models on the endpoint. With this launch, customers gain visibility into how much limits they have on the bedrock-mantle endpoint and can proactively plan for production scale. To get started, open the AWS Service Quotas console, choose Amazon Bedrock, and search for «Bedrock Mantle» to view your current quotas. To request an increase to any of these quotas, follow the standard Amazon Bedrock limit increase process. Service Quotas support for the bedrock-mantle endpoint is available in all AWS Regions where the endpoint is offered: US East (N. Virginia, Ohio), US West (Oregon), Asia Pacific (Mumbai, Tokyo, Sydney, Jakarta), Europe (Frankfurt, Ireland, London, Milan, Stockholm), and South America (São Paulo). To learn more, see Quotas for Amazon Bedrock.
We are pleased to announce general availability of Amazon EC2 P6-B200 instances in AWS US East (N. Virginia) on SageMaker notebook instances.
Amazon EC2 P6-B200 instances are powered by 8 NVIDIA Blackwell GPUs with 1440 GB of high-bandwidth GPU memory and 5th Generation Intel Xeon processors (Emerald Rapids). These instances deliver up to 2x better performance compared to P5en instances for AI training. Customers can use P6-B200 instances to interactively develop and fine-tune large foundation models, including LLMs, mixture of experts models, and multi-modal reasoning models. These instances enable efficient experimentation with larger models directly in JupyterLab or CodeEditor environments for generative AI applications such as enterprise copilots and content generation across text, images, and video.
We are pleased to announce general availability of Amazon EC2 P6-B200 instances in AWS US East (N. Virginia) on SageMaker notebook instances.
Amazon EC2 P6-B200 instances are powered by 8 NVIDIA Blackwell GPUs with 1440 GB of high-bandwidth GPU memory and 5th Generation Intel Xeon processors (Emerald Rapids). These instances deliver up to 2x better performance compared to P5en instances for AI training. Customers can use P6-B200 instances to interactively develop and fine-tune large foundation models, including LLMs, mixture of experts models, and multi-modal reasoning models. These instances enable efficient experimentation with larger models directly in JupyterLab or CodeEditor environments for generative AI applications such as enterprise copilots and content generation across text, images, and video.
Visit developer guides for instructions on setting up and using JupyterLab and CodeEditor applications on SageMaker Studio and SageMaker notebook instances.
AWS Glue now offers large and memory-optimized workers in the AWS Europe (Spain) Region, giving customers in this region more power to handle complex data processing workloads. The new additions include two general compute workers (G.12X and G.16X) as well as four memory-optimized workers (R.1X, R.2X, R.4X, and R.8X). With these options, you can now tackle more complex transforms, aggregations, joins, and queries while processing higher volumes of data quickly using AWS Glue.
The G.12X and G.16X workers extend the existing G worker lineup with additional compute, memory, and storage which makes them ideal for large, resource-intensive workloads. The R-series workers (R.1X, R.2X, R.4X, and R.8X) offer double the memory of their G counterparts, making them well-suited for memory-intensive Spark operations such as caching, shuffling, and aggregating. You can select any of these worker types through AWS Glue Studio, using notebooks or Visual ETL, or programmatically via the Glue Job APIs.
For more information on these worker types and AWS Regions where they are available, visit the AWS Glue documentation.
AWS Glue now offers large and memory-optimized workers in the AWS Europe (Spain) Region, giving customers in this region more power to handle complex data processing workloads. The new additions include two general compute workers (G.12X and G.16X) as well as four memory-optimized workers (R.1X, R.2X, R.4X, and R.8X). With these options, you can now tackle more complex transforms, aggregations, joins, and queries while processing higher volumes of data quickly using AWS Glue. The G.12X and G.16X workers extend the existing G worker lineup with additional compute, memory, and storage which makes them ideal for large, resource-intensive workloads. The R-series workers (R.1X, R.2X, R.4X, and R.8X) offer double the memory of their G counterparts, making them well-suited for memory-intensive Spark operations such as caching, shuffling, and aggregating. You can select any of these worker types through AWS Glue Studio, using notebooks or Visual ETL, or programmatically via the Glue Job APIs. For more information on these worker types and AWS Regions where they are available, visit the AWS Glue documentation.
